Landslides (2009) 6:225230

DOI 10.1007/s10346-009-0165-4
Received: 13 December 2008
Accepted: 21 July 2009
Published online: 12 August 2009
Springer-Verlag 2009
Markus Stoffel
.
Michelle Bollschweiler
Tree-ring reconstruction of past debris flows based
on a small number of samplespossibilities
and limitations
Abstract Tree-ring analyses have often been used in the past for
the reconstruction of spatiotemporal patterns of previous debris-
flow activity, often yielding very precise and extensive data for
torrents where information on former activity was largely missing.
Unless dendrogeomorphology is slated for multimillion Euro
developments, the large sets of tree-ring series that are usually used
in these studies render analysis time-consuming and not
necessarily very cost-effective. In this study, we present results
on past debris flows obtained with 35 Larix decidua Mill. trees
growing on the cone of the Torrent de Ptrey (Zinal, Valais, Swiss
Alps). It is concluded that studies based on a limited number of
samples may yield valuable data on past events, but that the
reconstructed frequency remains widely incomplete and indica-
tions on the spatial aspects of past events are only fragmentary.
Keywords Debris flow
.
Dendrogeomorphology
.
Tree-rings
.
Frequency
.
Valais Alps
.
Switzerland
Introduction
Detailed knowledge on the temporal and spatial incidence of
debris flows is of crucial importance as soon as anthropogenic
activity interferes with potentially hazardous processes (Bloetzer et
al. 1998). In this sense, data on the previous occurrence of events
are of crucial importance for the assessment of hazards and risks as
well as for the design of, e.g., torrent control works or retention
basins (Rickenmann 1999). As a consequence, attention has been
directed toward the analysis and documentation of process
dynamics in the aftermath of the widespread flooding and
debris-flow events in July and August 1987 (Rickenmann and
Zimmermann 1993), September 1993 (Rthlisberger 1994), October
2000 (BWG 2002), or August 2005 (Bezzola and Hegg 2007),
affecting large parts of the Swiss Alps and each of them causing
damage of several hundred million Euros.
Only a few torrents and gullies have been monitored over
sufficiently long periods of the past, and archival records on events
remain incomplete, despite recognition that debris flows are more
erosive and have a higher hazard potential than floods of
comparable return periods (Pierson 1980; de Scally and Owens
2005). Due to the general lack of documented events, tree-ring
records have repeatedly been used in the past to obtain
information on previous events (i.e., Strunk 1997; Baumann and
Kaiser 1999; Bollschweiler and Stoffel 2007), gain knowledge on
spatiotemporal patterns of former debris-flow activity on cones
(Bollschweiler et al. 2007, 2008a), investigate meteorological
conditions triggering debris flows (Stoffel et al. 2005a; Pelfini and
Santilli 2008), or predict the impact of changing climatic
conditions on past as well as potential future events (Stoffel and
Beniston 2006; Stoffel et al. 2008). In all these studies, several
hundreds to up to more than a thousand trees have been analyzed
and very extensive and complete information was obtained on the
local debris-flow history of the past few centuries.
While yielding important and large amounts of data on
previous events, the large sets of tree-ring series used in these
studies also render analysis time-consuming and not very cost-
effective for some applications. In this paper, we present results on
spatiotemporal patterns of past debris flows obtained from 35
Larix decidua Mill. growing on the cone of the Torrent de Ptrey
(Zinal, Valais, Swiss Alps) and highlight possibilities and limita-
tions of the approach.
Study area
The area investigated in this study is the cone of the Torrent de
Ptrey, located in the southern part of the village of Zinal (Valais,
Swiss Alps, 4608 N/738 E; Figs. 1 and 2). The catchment area of
the Ptrey torrent covers 1.6 km
2
, and the length of the primary
channel totals 1.8 km. The cone itself extends from approximately
1,680 to 1,800 m a.s.l., has a cone area of 4.2 ha, and is covered with
a forest composed of European larch (L. decidua Mill.) around the
currently active channel. The distal portions of the cone are not
forested and have served as pasture land over centuries, before
increased economic and touristic demands have led to the
construction of residential and vacation homes since the 1960s.
Debris-flow sediment predominantly originates from the
Glacier Bonnard rock glacier below the Diablons des Dames
(3,609 m a.s.l.; Fig. 1). Contemporary horizontal movements of the
rock-glacier body vary between 0.3 and 1 m year
1
(Delaloye et al.
2007), thus providing the Ptrey as well as the neighboring Tracuit
torrents with sediments belonging to the Cretaceous schists of the
Tsat nappe (Penninic units; Labhart 2004). The high elevation
of the departure zone (>2,400 m a.s.l.) and the presence of
contemporary permafrost restrict the release of debris flows to a
few months during the summer and early autumn. Debris-flow
material also originates from an active landslide located SSWof the
torrent between 1,800 and 2,000 m a.s.l. (Fig. 1).
Evidence for past debris-flow activity is scarce for the Torrent
de Ptrey and archives yield data on only five events between 1929
and today (1929, 1932, 5 July 1987, 20 June 1998, 4 August 2003).
Data are also largely missing for the neighboring Torrent de Tracuit
(3 July 1987). Only two major flood episodes are noted for La
Naviscence (24 September 1993, 14/15 October 2000), which is the
main river of the Val de Zinal valley (see Fig. 1). After a reaction to
the recent debris-flow activity, a dam and a retention basin (Fig. 1)
have been constructed on the lower part of the cone.
With respect to federal regulation, all Swiss communities need
to complete debris-flow and flooding hazard maps by 2011. Hazard
zoning will have important consequences on land-use planning
and on the determination of priorities for (non-)constructive
countermeasures, and consequently, hazard maps need to be based
Original Paper
Landslides 6 (2009) 225
on reliable input data. In regions where archival data are
nonexisting or incomplete, the history of past incidences can be
reconstructed with tree-ring analysis and through the use of
debris-flow models.
Materials and methods
Geomorphic mapping of debris-flow features
Analysis of past debris-flow activity begins with detailed mapping
of all features associated with past events, such as lobes, leves, or
abandoned flow paths in a scale of 1:1,000. Features and deposits
originating from other geomorphic processes or anthropogenic
activity are also mapped to avoid erroneous dating of debris-flow
events. Due to the presence of a relatively dense forest cover, global
positioning system cannot normally be used on the cone, which is
why geomorphic mapping is executed with a tape, compass, and
inclinometer.
Sampling design
On the Ptrey cone covering approximately 4 ha, only a small
minority of the predominant European larch trees show visible
growth defects (GD) related to past debris-flow activity (e.g.,
corrasion scars, tilted stems, partially buried trunks, decapitation,
partial destruction of root mass; see Stoffel and Bollschweiler 2008,
2009 for details). Based on the detailed geomorphic map and on an
outer inspection of the stem surface, we extracted at least two cores
per disturbed tree using Suunto increment borers, one in the flow
direction of past debris flows and the other on the opposite side of
the trunk (maximum core length, 40 cm6 mm). Cores were
preferably sampled at the height of the visible damage or within the
segment of the stem tilted during past events.
In addition to the disturbed trees sampled on the cone, we
selected undisturbed reference trees from a forest stand located
northeast of the cone, indicated with an asterisk in Fig. 1. For every
single reference tree, two cores per tree were extracted parallel to
the slope direction. In total, 64 trees were sampled (128 increment
cores): 35 trees (70 cores) from the debris-flow cone and 29 trees
(58 cores) from undisturbed reference sites. In contrast to the
disturbed trees, increment cores of the reference trees were
extracted at breast height (130 cm).
Data recorded for each tree sampled included (a) determination
of its position within leves, flow channels, or on deposits, (b)
sketch and position of visible GD, (c) sampling position of cores
with respect to the stem surface, (d) tree diameter at breast height
(130 cm), and (e) data on neighboring trees.
Reconstruction of past debris-flow activity
Samples were analyzed and data processed following the standard
procedures described in Brker (2002). Single steps of sample
analysis included surface preparation, skeleton plots as well as
ring-width measurements using digital LINTAB positioning tables
connected to a Leica stereomicroscope and TSAP 3.0 (Time Series
Analysis and Presentation) software (Rinntech 2008). Growth
curves of the disturbed samples were then crossdated with the
corresponding reference chronology constructed from undis-
turbed L. decidua trees, in order to separate insect attacks or
climatically driven fluctuations in tree growth from GD caused by
debris flows (Cook and Kairiukstis 1990).
Fig. 1 The Torrent de Ptrey (Zinal, Valais Alps, Switzerland) rises from its source
at ~2,600 m a.s.l. and passes through a forested cone, before converging with the
La Navisence river (1,660 m a.s.l.). The asterisk indicates the forest stand where
reference trees have been selected
Fig. 2 Debris-flow cone of the Torrent de Ptrey (1,6601,800 m a.s.l.; 4.2 ha) as
seen from the departure zone
Original Paper
Landslides 6 (2009) 226
Increment curves were then used to determine the initiation of
abrupt growth reduction or recovery (McAuliffe et al. 2006). In the
case of tilted stems, both the appearance of the cells (i.e., structure
of the reaction wood cells) and the growth curve data were
analyzed (Fantucci and Sorriso-Valvo 1999; Stoffel et al. 2006).
Finally, the cores were visually inspected to identify further signs of
past debris-flow activity in the form of callus tissue overgrowing
abrasion scars or tangential rows of traumatic resin ducts formed
following cambium damage (Bollschweiler et al. 2008b; Stoffel
2008). Figure 3 gives an overview on the visible GD that we were
looking for in the field and on how they influence yearly increment
rates and wood anatomy of L. decidua (i.e., both visibly on the
cross sections as well as in the growth curves).
Due to the small number of trees available for analysis, we
distinguish between certain and possible events. In the results
section, these two types of events are given with solid and dashed
lines, respectively. A reconstructed debris-flow event is considered
sure if (a) at least two trees show severe GD at the same time of the
year and (b) if the position of affected trees in the field makes
sense. In contrast, an event was only considered possible if only
one tree showed severe and several others exhibited simultaneous
yet less traumatic reactions.
Results
Geomorphology of the forested cone
The features and deposits inventoried in the study area covering
4.2 ha included 18 lobes, seven leves, and the currently used
channel. In contrast, there were no signs of well-developed, but
currently abandoned, channels observable in the field. In the
lowest part of the cone, signs of former debris-flow activity are
inexistent and the terrain completely remodeled through the dam
Fig. 3 Detailed map of the study site with debris-flow lobes and deposits, the position of sampled trees as well as human constructions (roads, tracks buildings, and dams)
Landslides 6 (2009) 227
construction works. Figure 4 illustrates the different features
identified on the cone and provides indications on the extent of the
forest stand as well as the position of the trees sampled.
Growth disturbances in trees and reconstructed debris-flow activity
Analysis of the disturbed trees allowed reconstruction of 97
characteristic GD caused by passing debris-flow surges or the
deposition of material on the cone (Table 1). Signatures of past
incidences were mainly identified on the increment cores via
tangential rows of traumatic resin ducts (63%), reaction wood
(18%), or abrupt growth reductions (15%). Growth recovery or
injuries were, in contrast, only occasionally found in the tree-ring
series. In total, the analysis of signatures occurring simultaneously
in different trees on the cone allowed the reconstruction of 22
debris-flow events covering the last 145 years, with nine being
considered possible and 13 defined as sure events. Figure 5 provides
the reconstructed frequency of debris flows between AD 1862 and
today.
Based on the spatial distribution of trees showing GD to the
same event, it was also possible to characterize the spatial spread of
surges during individual events, as shown with two examples in
Fig. 6. Figure 6a illustrates the GD associated with a debris flow
that occurred in 1959, when debris-flow material apparently left the
main channel at the level of the bridge and deposited several debris
lobes in the forest. In 1896 (Fig. 6b), only 11 trees were present for
analysis. Two of them are located in a debris-flow leve and show
tangential rows of traumatic resin ducts as a result of debris-flow
activity.
Discussion and conclusions
In the study we report here, 70 increment cores extracted from 35
living L. decidua Mill. trees allowed reconstruction of 97 growth
defects belonging to 22 debris-flow events since AD 1862. In
comparison to the five previously documented events, we were
able to extent the history of incidences in the Torrent de Ptrey
(Zinal, Valais, Swiss Alps) by more than 60 years back to AD 1862
and to augment the number of documented events.
This information is valuable to the local and regional
authorities (i.e., natural hazards and land-use planning autho-
rities) and the civil and hydraulic engineers being in charge of the
planning of passive debris-flow mitigation. In addition, the small
number of samples allowed reconstruction of previous activity and
a rough estimation on how often the torrent produced debris flows
within reasonable time and with limited effort.
Nonetheless, it is worthwhile to note that the reconstructed
frequency may only represent a minimum frequency of events for
the Ptrey torrent. All those surges having remained in the main
channel did not (necessarily) affect trees growing on the cone. As a
consequence, these events will not be identified in the tree-ring
series, unless the flowing material was denudating parts of the root
mass or tilting the stem axis of trees growing along the channels
edges. In addition and as a result of the comparably small number
of trees sampled on the Ptrey cone, it is also probable that even
some events that actually left the channel and flew over some parts
of the cone may have passed between the selected trees and did
not, therefore, disturb the 35 trees selected within this study. This
limitation may, however, not be significant for hazard and risk
Fig. 4 Evidence used to infer debris-
flow events from growth anomalies in
tree-ring records (adapted from Stoffel
et al. 2005b; Bollschweiler and Stoffel
2007)
Table 1 Debris-flow signatures identified in the 70 increment cores analyzed
Signature Number Percent
TRD 61 63
Wound 2 2
Reaction wood 17 18
Growth reduction 15 15
Growth increase 2 2
Total 97 100
TRD tangential rows of traumatic resin ducts
Original Paper
Landslides 6 (2009) 228
studies as it truncates the debris-flow frequency for the lower
magnitude events. Most prudent debris-flow mitigation measures
will rely on a high return period event, whose frequency is more
important that often occurring smaller events.
In addition, tree-ring-based dating of past debris-flow activity is
not normally based on one reaction in one single tree, but
determined through a quantitative (= minimum number of trees;
threshold exceedance; Butler et al. 1987) or semiquantitative
analysis (= number and spatial distribution of trees; Bollschweiler
et al. 2007; Stoffel et al. 2008) of trees showing GD in a particular
year. Hence, it will be more difficult to reduce uncertainty and to
accurately identify past debris-flow events when only a limited
number of trees are selected on the cone in general and in
individual sectors in particular.
Similarly, the validity of results obtained with a limited number
of samples is also reduced as soon as studies are focusing on the
spatial extent of debris-flow events or on the activity in currently
abandoned channels. Due to the limited sample depth, the number
of trees showing simultaneous reactions is often too small to track
the paths of previous events or to attribute lobate deposits to
specific events. Exceptions are always possible, as shown with the
two examples in Fig. 6. As a further consequence of lacking
information on the spread of debris-flow material on the cone or
on the deposition of lobes, it seems illusory to estimate the volume
of past events. Therefore, reliable frequencymagnitude relation-
ships are almost impossible to achieve even on sites where
extensive datasets have been created through the tree-ring-based
analysis of important numbers of trees and the subsequent dating
of deposits on the current-day surface (Strunk 1986; Stoffel et al.
2005a, 2008).
Based on the results and the above considerations, it appears
possible to characterize past debris-flow activity in and to provide
a minimum frequency for the Torrent de Ptrey with a very
limited number of trees sampled. The limited amount of time
invested renders analysis more cost-effective but may, as a
consequence, only provide a rather incomplete dataset and
exclusively data on the past existence of debris flows. Data on
the spread of material on the cone, the activity in currently
abandoned channels, the age of deposits, or the magnitude of past
events cannot be obtained with this approach.
Fig. 6 Spatial representation of trees
affected during debris-flow activity in a
1959 and b 1896 (red dots). Trees living
at the time of the event but remaining
unaffected by the event are repre-
sented with gray dots
Fig. 5 Debris-flow frequency showing
22 events reconstructed since AD 1862
from the tree-ring series. Solid lines
indicate sure events where at least two
trees showed severe growth distur-
bances. If only one tree showed severe
and several others exhibited simulta-
neous yet less traumatic growth dis-
turbances, reactions were considered
the result of a possible event (dashed
lines)
Landslides 6 (2009) 229
Acknowledgments
This paper is dedicated to our friend Grgoire Devaud (19742006),
who passed away while climbing in his beloved mountains. Work
has been undertaken in the context of the RUFINE project financed
by the Swiss Federal Office for Environment as well as the Service
Fort et Paysage and the Service des Routes et Cours dEau (both
canton of Valais). We offer our sincere thanks to the anonymous
reviewers for their insightful comments.
References
Baumann F, Kaiser KF (1999) The Multetta debris fan, Eastern Swiss Alps: a 500-year
debris flow chronology. Arct Antarct Alp Res 31:128134
Bezzola GR, Hegg C (eds) (2007) Ereignisanalyse Hochwasser 2005. Teil 1Prozesse,
Schden und erste Einordnung. Bundesamt fr Umwelt BAFU, Eidgenssische
Forschungsanstalt WSL. Umweltwissen 0707, Bern
Bloetzer W, Egli T, Petrascheck A, Sauter J, Stoffel M (1998) Klimanderungen und
Naturgefahren in der RaumplanungMethodische Anstze und Fallbeispiele.
Synthesebericht NFP31. vdf Hochschulverlag, Zrich
Bollschweiler M, Stoffel M (2007) Debris flows on forested conesreconstruction and
comparison of frequencies in two catchments in Val Ferret, Switzerland. Nat Hazards
Earth Syst Sci 7:207218
Bollschweiler M, Stoffel M, Ehmisch M, Monbaron M (2007) Reconstructing spatio-
temporal patterns of debris-flow activity using dendrogeomorphological methods.
Geomorphology 87(4):337351
Bollschweiler M, Stoffel M, Schneuwly DM (2008a) Dynamics in debris-flow activity on a
forested conea case study using different dendrochronological approaches. Catena
72(1):6778
Bollschweiler M, Stoffel M, Schneuwly DM, Bourqui K (2008b) Where do traumatic resin
ducts occur in Larix decidua trees that have been impacted by debris flows? Tree
Physiol 28:255263
Brker OU (2002) Measuring and data processing in tree-ring researcha
methodological introduction. Dendrochronologia 20(12):203216
Butler DR, Malanson GP, Oelfke JG (1987) Tree-ring analysis and natural hazard
chronologies: minimum sample size and index values. Prof Geogr 39:4147
BWG (Bundesamt fr Wasser und Geologie) (ed) (2002) Hochwasser 2000Les crues
2000. Berichte des Bundesamtes fr Wasser und Geologie, Serie Wasser 2:1248
Cook ER, Kairiukstis LA (1990) Methods of dendrochronologyapplications in the
environmental sciences. Kluwer, London
Delaloye R, Bardou E, Hagin C (2007) Glacier Bonnard. Mouvements du glacier rocheux.
Troisime campagne de mesure. Unpublished intermediate report. University of
Fribourg, Idalp Srl and Geosat AG, Fribourg and Sion
de Scally FA, Owens IF (2005) Depositional processes and particle characteristics on fans
in the Southern Alps, New Zealand. Geomorphology 69:4656
Fantucci R, Sorriso-Valvo M (1999) Dendrogeomorphological analysis of a slope near
Lago, Calabria (Italy). Geomorphology 30:165174
Labhart TP (2004) Geologie der Schweiz, 6th edn. Ott, Thun
McAuliffe JR, Scuderi LA, McFadden LD (2006) Tree-ring record of hillslope erosion and
valley floor dynamics: landscape responses to climate variation during the last 400 yr
in the Colorado Plateau, northeastern Arizona. Glob Planet Change 50(34):184201
Pelfini M, Santilli M (2008) Frequency of debris flows and their relation with
precipitation: a case study in the central Alps, Italy. Geomorphology 101:721730
Pierson TC (1980) Erosion and deposition by debris flows at Mt. Thomas, north
Canterbury, New Zealand. Earth Surf Processes Landf 6:227247
Rickenmann D (1999) Empirical relationships for debris flows. Nat Hazards 19:4777
Rickenmann D, Zimmermann M (1993) The 1987 debris flows in Switzerland:
documentation and analysis. Geomorphology 8:175189
Rinntech 2008. LINTABprecision ring by ring. http://www.rinntech.com/Products/
Lintab.htm.
Rthlisberger G (1994) Unwetterschden in der Schweiz im Jahre 1993. Wasser Energie
Luft 86(12):18
Stoffel M (2008) Dating past geomorphic processes with tangential rows of traumatic
resin ducts. Dendrochronologia 26:5360
Stoffel M, Beniston M (2006) On the incidence of debris flows from the early Little Ice
Age to a future greenhouse climate: a case study from the Swiss Alps. Geophysical
Research Letters 33:L16404
Stoffel M, Bollschweiler M (2008) Tree-ring analysis in natural hazards researchan
overview. Nat Hazards Earth Syst Sci 8:187202
Stoffel M, Bollschweiler M (2009) What tree rings can tell about earth-surface processes.
Teaching the principles of dendrogeomorphology. Geography Compass 3:10131037
Stoffel M, Bollschweiler M, Hassler GR (2006) Differentiating events on a cone influenced
by debris-flow and snow avalanche activitya dendrogeomorphological approach.
Earth Surf Processes Landf 31(11):14241437
Stoffel M, Conus D, Grichting MA, Livre I, Matre G (2008) Unraveling the patterns of
late Holocene debris-flow activity on a cone in the central Swiss Alps: chronology,
environment and implications for the future. Glob Planet Change 60:222234
Stoffel M, Livre I, Conus D, Grichting MA, Raetzo H, Grtner HW, Monbaron M (2005a)
400 years of debris-flow activity and triggering weather conditions: Ritigraben,
Valais, Switzerland. Arct Antarct Alp Res 37(3):387395
Stoffel M, Schneuwly D, Bollschweiler M, Livre I, Delaloye R, Myint M, Monbaron M
(2005b) Analyzing rockfall activity (16002002) in a protection foresta case study
using dendrogeomorphology. Geomorphology 68(34):224241
Strunk H (1997) Dating of geomorphological processes using dendrogeomorphological
methods. Catena 31:137151
Strunk H (1986) Episodische Materialverlagerungen und die Fragwrdigkeit von
Bilanzierungen. Darmstdter Geographische Studien 7:4557
M. Stoffel ())
.
M. Bollschweiler
Laboratory of Dendrogeomorphology, Institute of Geological Sciences,
University of Berne,
Baltzerstrasse 1+3,
3012 Berne, Switzerland
e-mail: markus.stoffel@dendrolab.ch
M. Stoffel
.
M. Bollschweiler
Climatic Change and Climate Impacts Group, Environmental Sciences,
University of Geneva,
chemin de Drize 7,
1227 Carouge, Switzerland
M. Stoffel
.
M. Bollschweiler
Department of Geosciences, Geography, University of Fribourg,
chemin du Muse 4,
1700 Fribourg, Switzerland
Original Paper
Landslides 6 (2009) 230